基于模拟退火遗传算法的渠系配水优化编组模型研究  

Optimal water delivery marshalling model based on simulated annealing genetic algorithm

作　　者：赵文举[1] 马孝义[1] 张建兴[1] 朱亚磊[1] 

机构地区：[1]西北农林科技大学水利与建筑工程学院,陕西杨凌712100

出　　处：《水力发电学报》2009年第5期210-214,113,共6页Journal of Hydroelectric Engineering

基　　金：国家自然科学基金项目(50879072;50479052);国家科技支撑计划课题(2006BAD11B04)

摘　　要：科学的渠系配水决策可减少渠系输水过程中的渗水损失,提高渠系水利用率。本文在分析渠系配水优化编组模型特点的基础上,针对基于固定惩罚函数遗传算法的模型求解方法中存在的量纲难以统一、搜索精度不高和过早收敛到局部最优解的问题,引入动态罚函数来处理约束条件,提出了基于模拟退火遗传算法的模型求解方法。通过灌区实际配水编组决策应用实例验证表明,本文方法收敛速度快,计算精度高,确定的下级渠道配水时间搭配合理,上级配水流量均匀,具有较高的工程应用价值。An optimal schedule of canals water delivery can reduce the water seepage loss and improve the water use efficiency of irrigation. The optimal schedule model is subjected to many constraints, however, which causes difficulty of the traditional genetic algorithm in seeking a solution. This algorithm easily converges to a local optimal solution and leads to a poor accuracy, because it uses the fixed-penalty functions to represent the constraints of different dimension. To overcome such a problem in the analysis of water delivery scheduling, this paper describes a method of adopting the dynamic penalty functions for model constraints and the simulated annealing genetic algorithm. Application to a practical irrigation canal system and comparison with other two methods shows that this highly stable method can improve the convergence and the solution accuracy. When the method is used for water delivery scheduling, the lateral delivery time and the upper canal flow rates can be better predicted, which often results in a much shorter water distribution time and a smaller seepage loss. This method is of great value to the practical irrigation canal system management.

关 键 词：灌溉渠系 配水编组 优化 模拟退火遗传算法 动态罚函数 

分 类 号：S274.2[农业科学—农业水土工程] TF311.52[农业科学—农业工程]